STAT 1000 Lecture Notes - Lecture 5: Scatter Plot, Confounding, Linear Equation
Document Summary
Correlation: a stat technique used to measure and describe a relationship between two quant variables quantitatively. Usualy the two variables are simply observed as they exist in nature, no attempt to manipulate variables. Requires two scores for each individual, the pair of score can be listed in a table or presented in a scatterplot. Properties of correlation: special values for r. r alwaus between 1 and -1. Correlation not affected by roles of explanatory response. Ex: plot heigh on x, weigh on y, or vice versa, still see positive correlation and same r value. Also not affected by change in units, if graph pounds vs kilograms, r stays the same. Problems with correlations: outliers can inflate or deflate correlation. Helps get rid of negatives number (-1) in negative correlations. Measures the gain in accuracy that is obtained from using the correlation for prediction. Measures the proportion of the variance in one variable explained by the other variable.